TruMan: Trust Management for Vehicular Networks

By integrating processors and wireless communication units into vehicles, it is possible to create a vehicular ad-hoc network (VANET), in which cars share data amongst themselves in order to cooperate and make roads safer and more efficient. A decentralized ad-hoc solution, which does not rely on previously existing infrastructure, Internet connection or server availability, is preferred so the message delivery latency is as short as possible in the case of life-critical situations. However, as it is the case with most new technologies, VANETs will be a prime target for attacks performed by malicious users, who may benefit from affecting traffic conditions. In order to avoid such attacks, one important feature for vehicular networks is trust management, which allows nodes to filter incoming messages according to previously established trust values assigned to other nodes. To generate these trust values, nodes use information acquired from past interactions. Nodes which frequently share false or irrelevant data must have lower trust values than the ones which appear to be reliable. This work proposes TruMan, a trust management model in the context of daily commutes, utilizing the Working Day Movement Model as a basis for node mobility. The results prove to be accurate, detecting nearly all malicious nodes with very few false positives when they constitute up to 50% of the network. The model is also very efficient thanks to the low complexity of the algorithm constituting the trust model.